Mobile delay-tolerant sensor networks are becoming increasingly important because of their ability to deliver long periods of fine-grained sensing over a wide area with a small number of nodes. A key challenge in these systems, however, is that nodes are extremely energy constrained since they must be small, lightweight, and function autonomously for months at a time. This problem is compounded by the fact that mobile nodes demand radios with relatively long ranges to maximize the effectiveness of short, infrequent communication periods.
This presentation will introduce my dissertation on energy conservation techniques for these networks. The talk will primarily focus on a family of lossless compression algorithms tailored to sensor networks. These algorithms include a novel LZW variant that exploits characteristic patterns of sensor data to reduce energy consumption by more than 40% as well as further data transforms that can take advantage of the structure of the data to decrease energy consumption by nearly a factor of three.
Then, this presentation will briefly introduce a data abstraction layer for mobile sensor networks that reorganizes data from the network's viewpoint. This organization facilitates the development of services for data identification, search, and reduction, which combine to save energy by making communications more efficient.